B-RAD 多模态检索增强乳腺超声诊断框架在七国队列与阅片研究中的验证
Interpretable multimodal retrieval augmented diagnosis for breast ultrasound with multinational clinical validation and reader study
来自首尔大学与国民大学团队在 npj Digital Medicine 发表的开放获取论文提出 B-RAD 框架,将活检分诊准确率提升至 0.952 AUROC,并在 11 个来自 7 国共计 8311 张图像的队列与 4 名阅片者研究中显示可改善间阅片者一致性、减少恶性病变漏诊。
在七个国家八个队列共 8311 张乳腺超声图像上验证,BI-RADS 活检分诊 AUROC 达 0.952,多模态检索辅助让四名阅片者间一致性由中度升至显著并降低漏诊。
- Article
- Open access
- Published: 03 October 2026
- Han Jang ORCID: orcid.org/0009-0002-4719-31801,2 na1,
- Subin An ORCID: orcid.org/0009-0007-2921-22843 na1,
- Sunghyun Jung2,
- Hye Ji2,
- Miran Kim4,
- Soochahn Lee3,
- Kyu Sung Choi2,5,6 &
- …
- Su Min Ha2,5
npj Digital Medicine (2026) Cite this article
We’re sharing this article early to provide faster access to peer-reviewed, accepted research. It is citable and carries a permanent DOI. This version is subject to further edits and will be replaced automatically by the final Version of Record. All legal disclaimers apply.
Abstract
Breast ultrasound is central to breast cancer diagnosis but suffers from substantial inter-observer variability in Breast Imaging Reporting and Data System (BI-RADS) assessment, and existing deep learning models remain opaque. Multimodal large language models promise interpretable report-based reasoning, yet scarce paired image-report data yields weak visual-semantic alignment. We propose B-RAD, a BI-RADS-aware retrieval-augmented diagnosis framework that mirrors the radiologist workflow of retrieving exemplars, localizing lesions, and determining the BI-RADS category. Its retrieval module learns BI-RADS-aware alignment from unpaired data by jointly minimizing cross-modal mismatch and ordinal prediction error. Retrieved exemplars guide few-shot detection of the lesion together with its margin and posterior acoustic features, and the detected region prompts a segmentation model whose mask drives training-free foveal attention for classification. We validated B-RAD on eleven cohorts from seven countries totaling 8311 images, spanning internal validation and three external cohorts including an independent institutional cohort. The full pipeline reached a biopsy triage AUROC of 0.952 on that institutional cohort without fine-tuning, outperforming existing vision-language models. In a four-reader study, B-RAD assistance improved accuracy, raised inter-reader agreement from moderate to substantial, and reduced missed malignancies across all readers. These findings show that retrieval-augmented diagnosis can narrow the expertise gap and support accessible breast cancer screening in resource-limited settings.
Acknowledgements
We thank the creators of the publicly available breast ultrasound datasets used in this study. This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. RS-2026-25479661 (K.S.C.), No. RS-2025-00521972 (S.L.)); the Seoul National University Hospital (SNUH) Research Fund (No. 04-2024-0600; No. 04-2025-2060) (K.S.C.); the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI) grant funded by the Ministry of Health & Welfare (No. RS-2024-00439549) (K.S.C.); the Korean Society of Magnetic Resonance in Medicine (KSMRM) Research Grant (No. 2024-R00) (K.S.C.); the Research Program 2025 funded by the Seoul National University College of Medicine Research Foundation (No. 800-20250240) (K.S.C.); and the Institute of Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (MSIT) (No. RS-2025-02219317, AI Star Fellowship, Kookmin University) (S.L.). This research was also supported by the Advanced GPU Utilization Support Program funded by the Government of the Republic of Korea (Ministry of Science and ICT) (K.S.C.). This work was partly supported by IITP through the AI Computing Support Project for R&D, funded by the Korea government (MSIT) (No. RS-2026-25505492, High-Performance Research AI Computing Infrastructure Support at the 2 PFLOPS Scale) (K.S.C.).
Author information
Author notes
These authors contributed equally: Han Jang, Subin An.
Authors and Affiliations
Interdisciplinary Program in Bioengineering, College of Engineering, Seoul National University, Seoul, Republic of Korea
Han Jang
Department of Radiology, Seoul National University Hospital, Seoul, Republic of Korea
Han Jang, Sunghyun Jung, Hye Ji, Kyu Sung Choi & Su Min Ha
School of Electrical Engineering, College of Engineering, Kookmin University, Seoul, Republic of Korea
Subin An & Soochahn Lee
Department of Radiology, Kyungpook National University Chilgok Hospital, Daegu, Republic of Korea
Miran Kim
Department of Radiology, Seoul National University College of Medicine, Seoul, Republic of Korea
Kyu Sung Choi & Su Min Ha
Healthcare AI Research Institute, Seoul National University Hospital, Seoul, Republic of Korea
Kyu Sung Choi
Authors
- Han Jang
- Subin An
- Sunghyun Jung
- Hye Ji
- Miran Kim
- Soochahn Lee
- Kyu Sung Choi
- Su Min Ha
Corresponding authors
Correspondence to Soochahn Lee or Kyu Sung Choi.
Ethics declarations
Competing interests
H.J., S.A., S.L., and K.S.C. are listed as inventors on a pending Korean patent application (Application No. 10-2026-0096895), filed by the Kookmin University Industry-Academic Cooperation Foundation and Seoul National University Hospital. This pending patent covers the end-to-end B-RAD pipeline described in this article, from multimodal retrieval to lesion localization and BI-RADS classification. The remaining authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Jang, H., An, S., Jung, S. et al. Interpretable multimodal retrieval augmented diagnosis for breast ultrasound with multinational clinical validation and reader study. npj Digit. Med. (2026). https://doi.org/10.1038/s41746-026-03272-3
Received: 02 April 2026
Accepted: 09 September 2026
Published: 03 October 2026
DOI: https://doi.org/10.1038/s41746-026-03272-3
来源:Nature 健康护理主题 · nature.com